| # Publish to Hugging Face | |
| Use the current `hf` CLI, not the deprecated `huggingface-cli` command. | |
| If `hf` is not available in the active environment, update the Hub package: | |
| ```bash | |
| python -m pip install -U huggingface_hub | |
| ``` | |
| Log in: | |
| ```bash | |
| hf auth login | |
| ``` | |
| Create the model repo. Keep it private while you inspect the uploaded card and | |
| files: | |
| ```bash | |
| hf repos create YOUR_USERNAME/Anima-Lightning --repo-type model --private --exist-ok | |
| ``` | |
| Upload this folder to the root of the model repo: | |
| ```bash | |
| hf upload YOUR_USERNAME/Anima-Lightning . . --repo-type model --commit-message "Upload Anima-Lightning distilled model" | |
| ``` | |
| After upload, check: | |
| - `README.md` renders with the `distilled_generation_config.json` runtime note. | |
| - `LICENSE.md` and `NOTICE.md` are present. | |
| - The model files remain in the same Diffusers subfolders: `transformer/`, | |
| `text_encoder/`, `text_conditioner/`, `vae/`, `scheduler/`, `tokenizer/`, | |
| and `t5_tokenizer/`. | |
| - The repo can stay private until you are ready to publish it. | |